At-home ketamine's entire evidence base came from the companies selling it

Ketamine for psychiatric conditions has grown into a multibillion-dollar industry in the United States, organised around three delivery models: in-person clinics offering off-label racemic ketamine; clinic visits for esketamine, which is approved; and telehealth services prescribing off-label racemic ketamine [s4]. About half the market consists of telehealth using off-label prescribing [s4].

That last model — a prescription issued remotely, a drug taken at home, no clinician in the room — is the one with the least evidence behind it, and this month a systematic review established just how little.

Three studies

Researchers publishing in Australasian Psychiatry followed PRISMA 2020 and searched for randomised or observational studies of adults receiving at-home, telehealth-supported ketamine for depression or anxiety, up to August 2025 [s1]. Their primary outcomes were a 50% or greater symptom reduction, and remission [s1].

They screened 3,857 records. Three met criteria [s1].

All three came from US commercial providers [s1]. The programmes differed from each other in dosing schedules, clinical staffing, supervision and safety monitoring [s1] — meaning the three studies do not describe a single intervention that could be pooled or compared.

Across up to 16,876 patients, follow-up outcome data were available for only a minority [s1].

What the numbers show, with the caveat attached

Among those with follow-up data, response ranged from 49.5% to 62.8% on the PHQ-9 and from 47.6% to 62.9% on the GAD-7 [s1]. Remission ranged from 20.7% to 32.6% and from 23.9% to 31.3% respectively [s1].

Those are not bad numbers. They are also, as the review states, produced entirely by provider-generated research, with all studies rated at critical overall risk of bias, and GRADE certainty for both effectiveness and safety judged very low [s1].

"Very low certainty" has a specific meaning in GRADE: the true effect is likely to be substantially different from the estimate. When response data exist for a minority of a self-selected commercial cohort, the patients missing from the denominator are systematically likely to be the ones who stopped, deteriorated, or did not improve.

The review's own conclusion is restrained. Telehealth-supported sublingual racemic ketamine was associated with symptom improvements in selected populations, but the evidence base is provider generated and at critical risk of bias, and its main contribution is to inform consideration of model components rather than to establish effectiveness [s1]. The authors note that the review protocol was not registered and the study was not funded [s1].

What the market looks like from outside

A separate cross-sectional study characterised the supply side directly. In 2025, researchers ran systematic web searches to identify clinics advertising prescription ketamine for psychiatric indications across the New York metropolitan area, then reviewed each clinic's public-facing website [s2].

They located 233 clinics [s2]. Over a third — 36.5% — prescribed ketamine for at-home use [s2]. Just over half, 51.5%, listed a medical doctor as part of their team [s2], and 42.9% advertised oral ketamine [s2].

Depression was the most commonly listed condition treated, at 94.0% [s2]. And 21.9% advertised ketamine as a treatment for substance use disorder [s2] — an indication with markedly thinner evidence than depression, marketed by roughly one clinic in five.

The multivariable model found that advertising ketamine for at-home use was more common among clinics advertising oral ketamine (adjusted prevalence ratio 4.10, 95% CI 2.20 to 7.60) and less common among clinics listing a medical doctor (aPR 0.54, 0.30 to 0.99) [s2].

That second association is the one to read twice. In this sample, the clinics sending a dissociative anaesthetic home with patients were less likely to have a physician on the listed team.

The study's limitation is that it examined only public-facing websites [s2], which describe what clinics advertise rather than what they do.

Where responsibility lands

An analysis in Psychiatric Services takes a wrongful death lawsuit against a telehealth ketamine provider as its point of departure and examines how combining off-label prescribing with on-demand delivery strains clinical safeguards [s3]. Given the limited oversight of many telehealth platforms and the drugs they prescribe, the authors argue, responsibility and liability for patient safety rest heavily on individual clinicians [s3]. Sustaining responsible care under those conditions requires support for robust clinical relationships, monitoring and accountability [s3].

The regulatory background is not accidental. COVID-19-era regulatory flexibilities enabled the expansion of direct-to-consumer telemedicine into mental health care [s3], and the legal changes behind the growth of telehealth-only providers came with differences in patient monitoring requirements [s4].

Off-label prescribing is legal and routine. The Journal of General Internal Medicine framing sets three questions a prescriber should answer before doing it: whether sound evidence exists to prescribe, whether the benefits outweigh the risks, and whether prescribing serves the patient's best interests [s4]. As a dissociative anaesthetic, ketamine raises particular concerns about whether its purpose and function meet the legal requirements for appropriate prescribing of a controlled substance [s4].

What to watch

The reviewed question is narrow and worth keeping narrow. It is not whether ketamine has antidepressant effects; it is whether the at-home telehealth model has been tested. On the systematic review's account, the entire published evidence base for that model consists of three commercially produced studies, all rated at critical risk of bias, with follow-up data for a minority of patients [s1] — supporting a delivery route that accounts for roughly half the US market [s4].

Whether independent trials of at-home protocols appear, and whether any regulator requires them before the model scales further, is the question the next two years will answer.

This article is informational and is not medical advice.